Deno

You are running a secure MCP server that lets AI assistants execute TypeScript and JavaScript code through Deno, with fine‑grained permissions and MCP‑compliant integration. This server enables you to safely run code in isolated Deno processes while controlling what the executed code can access.

How to use

Set up your MCP client to connect to the Deno MCP server using a stdio configuration. You start the server by running the command and arguments shown in the configuration examples, then you can issue MCP requests to execute code within the Deno runtime. You control what the executed code can access by selecting the appropriate startup permissions when you configure the client.

If you are using Claude Desktop, add a server entry so your client can send code execution requests to the Deno MCP server. The configuration below demonstrates the required shape and keys to reference when you wire up the client.

{
  "mcpServers": {
    "deno-executor": {
      "command": "deno",
      "args": [
        "run",
        "--allow-all",
        "jsr:@cong/mcp-deno"
      ]
    }
  }
}

How permissions shape what code can do

The server uses Deno’s permission system to bound what executed code can access. You configure those permissions at startup, so you can allow only the resources you trust.

For daily use, you might start with a conservative permission set and expand only when you need additional capabilities.

Examples of common permission controls you can apply include network access, file system read/write, and environment access.